Similar things can be said about text generators like ChatGPT. We’ve seen people use it to create a jumping on point for simple Python code, which is an interesting use of open-source scripts. However, for actual text… well, the best way I’ve seen it described is ‘confidently wrong’. For very basic stuff, you might get something competent, but anything else and it’s better to get a person involved.
Over the years, we’ve made great use of OpenCV in many Raspberry Pi projects – it’s a machine learning tool that processes image data to help make decisions. It’s used in robotics a lot with a Raspberry Pi Camera Module, with the late, great Formula Pi robots doing stellar work with it as they zipped around.
This use of AI is great. It’s doing something people can sort of do, but if you were doing these tasks it would be a bit menial. And the AI can do it a lot faster.
I’ve also seen people use image recognition technology with Raspberry Pi to create a Pokédex from the Pokémon anime – a device that recognises the eponymous monsters – and simply a ‘magic wand’ that identifies what it’s being pointed at. Both work to varying degrees of success. However, humans only know a limited amount of knowledge too.
We’re currently at a tipping point with a lot of AI tech. It’s getting very interesting and very powerful, but the technology should not be replacing human expression – especially when it’s copying other humans’ expressions without any context. The way it’s been used in computer vision is a good template for the future – as an aid and not a pseudo-replacement. Art is important expression after all.
With the new Camera Module 3 and its laundry list of new features, I can only imagine what we’ll be able to do with it that a person cannot.